At present, two Green's function-based methods represent the 'gold standard' for fully ab initio predictions of electronic excitations within the computational condensed-matter/solid state community: 1) the GW approximation is used to describe individual electron states and energies (quasiparticle properties); 2) the Bethe-Salpeter Equation is used to determine optical and excitonic properties (electron-hole interactions). After outlining the main ideas and physical approximations employed in each method, I will review algorithmic choices and computational costs, recent advances in this area, and hurdles regarding their parallel scaling. In addition, I will describe our collaborative effort as part of a recently funded NSF SI2-SSI software institute that brings together materials physicists and chemists with computer scientists to tackle the scientific, algorithmic, and software infrastructure challenges.

Ab initio excited state Green's function methods and their scaling on massively parallel architectures

248th National Meeting (2014)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.